Voltage to current converters which compensate for a temperature coefficient are well known in the art. For example, referring to FIG. 1, one version of a voltage to current converter (hereinafter circuit 10), described in Analysis and Design of Analog Integrated Circuits, 2nd Edition, by P. R. Grey and R. E. Meyer, (1977, 1980), has a reference voltage V.sub.REF applied to line 1 which is coupled to the positive input terminal of operational amplifier 2. The output line 3 of operational amplifier 2 is connected to the gate G of N-channel transistor 9. Feedback loop 6 couples the source S of transistor 9 to the negative input terminal of operational amplifier 2. Source S is also coupled to resistor R which is grounded. The output current I.sub.OUT is provided on line 5 which is coupled to the drain D of transistor 9. Voltage to current conversion is accomplished by maintaining reference voltage V.sub.REF across resistor R using operational amplifier 2. By definition, the voltage V.sub.REF on line 1, connected to the positive input of operational amplifier 2, will also appear at node 8. Circuit 10 will be independent of temperature only if resistor R and reference voltage V.sub.REF have the same temperature coefficient, i.e. are affected by temperature in a similar manner. Since, on silicon, the best resistor available has a temperature coefficient of approximately 0.1% per degree Celsius, it follows that V.sub.ref must have a similar temperature coefficient. In addition to the temperature coefficient, the value of resistor R and reference voltage V.sub.REF are also typically dependent on processing.
To compensate for the temperature coefficient of resistor R, voltage V.sub.REF must exhibit a temperature coefficient equal to that of resistor R. One way of achieving a variable reference voltage V.sub.REF which would compensate for the temperature coefficient of resistor R is through the use of a band-gap voltage reference circuit (not shown) connected to line 1. Band-gap voltage reference circuits require the use of bipolar transistors. However, in CMOS processing, bipolar transistors pose several disadvantages. First, CMOS processing typically provides "parasitic" (i.e. non-dedicated) bipolar devices, but their use requires an undesirable substrate current. Second, dedicated bipolar transistors require special processing which increases manufacturing cost and complexity.
Therefore, a need arises for a voltage to current converter which compensates for a temperature coefficient, and which can be implemented in full CMOS processing (i.e. without bipolar transistors).